Download Financial Leverage and the Leverage Effect

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts

Private equity wikipedia , lookup

Investment management wikipedia , lookup

Household debt wikipedia , lookup

Private equity secondary market wikipedia , lookup

Mark-to-market accounting wikipedia , lookup

Business valuation wikipedia , lookup

Greeks (finance) wikipedia , lookup

Private equity in the 1980s wikipedia , lookup

Stock trader wikipedia , lookup

Debt wikipedia , lookup

Stock valuation wikipedia , lookup

Beta (finance) wikipedia , lookup

Financialization wikipedia , lookup

Systemic risk wikipedia , lookup

Corporate finance wikipedia , lookup

Financial economics wikipedia , lookup

Transcript
Carnegie Mellon University
Research Showcase @ CMU
Tepper School of Business
3-2007
Financial Leverage and the Leverage Effect - A
Market and Firm Analysis
A. Cevdet Aydemir
Lehman Brothers
Michael Gallmeyer
Texas A & M University
Burton Hollifield
Carnegie Mellon University, [email protected]
Follow this and additional works at: http://repository.cmu.edu/tepper
Part of the Economic Policy Commons, and the Industrial Organization Commons
This Working Paper is brought to you for free and open access by Research Showcase @ CMU. It has been accepted for inclusion in Tepper School of
Business by an authorized administrator of Research Showcase @ CMU. For more information, please contact [email protected].
Financial Leverage and the Leverage Effect - A Market and Firm
Analysis ∗
A. Cevdet Aydemir
Lehman Brothers
Michael Gallmeyer†
Mays Business School
Texas A&M University
Burton Hollifield
Tepper School of Business
Carnegie Mellon University
Current Draft: March 2007
ABSTRACT
We quantify the effect of financial leverage on stock return volatility in a dynamic general
equilibrium economy with debt and equity claims. The effect of financial leverage is studied both
at a market and a firm level where the firm is exposed to both idiosyncratic and market risk.
In a benchmark economy with both a constant interest rate and constant price of risk, financial
leverage generates little variation in stock return volatility at the market level but significant
variation at the individual firm level. In an economy that generates time-variation in interest
rates and the price of risk, there is significant variation in stock return volatility at the market
and firm level. In such an economy, financial leverage has little effect on the dynamics of stock
return volatility at the market level. Financial leverage contributes more to the dynamics of
stock return volatility for a small firm.
Keywords: stock volatility, leverage effect, corporate debt, general equilibrium
JEL Classification: G10, G12
∗
We thank David Chapman, Rick Green, Antonio Mele, Bradley Paye, Chris Telmer, Jessica Wachter, and seminar participants at the 17th FEA Conference, the Adam Smith Asset Pricing Conference, the American Finance
Association, the Copenhagen School of Business, HEC Lausanne, the Norwegian School of Business in Oslo, the Oslo
School of Economics, the Second Annual Lone Star Finance seminar, and the Society of Economic Dynamics Meeting
for helpful comments.
†
Email: [email protected].
1. Introduction
The leverage effect is that stock volatility is negatively correlated to stock returns—stock volatility
tends to increase when stock prices drop. There are two common economic explanations for the
leverage effect. The first explanation is based on the relationship between volatility and expected
returns. When volatility rises, expected returns tend to increase, leading to a drop in the stock price.
As a consequence, volatility and stock returns are negatively correlated. The second explanation
is based on financial leverage. When stock prices fall, financial leverage increases, leading to an
increase in stock return volatility.
No consensus exists on the size of the effects of financial leverage on stock volatility. Empirical
studies that quantify the effect have produced mixed results, but these studies face several difficulties. Any study of the effect of financial leverage on volatility should use market debt valuations,
which are difficult to obtain in practice. The empirical literature also lacks a theoretical benchmark
for asset volatilities consistent with many features of asset pricing data in the presence of both debt
and equity claims. We provide such a benchmark.
The hypothesis that financial leverage can explain the leverage effect was first discussed by
Black (1976) and Christie (1982). Christie (1982) provides empirical evidence—based on a sample
of large firms—for the negative relationship between stock returns and volatility induced by financial
leverage. Duffee (1995) argues that such a relationship does not hold when small firms are also
included in the sample. Schwert (1989) shows empirically that financial leverage cannot fully
account for the observed variation in market volatility. Figlewski and Wang (2000) document a
stronger leverage effect in down markets than in up markets. Campbell and Hentschel (1992)
and Bekaert and Wu (2000) provide econometric models of asymmetric volatility. None of these
empirical studies use market debt values to compute financial leverage.
Wu (2001) builds a partial equilibrium model to study the sources of asymmetric volatility.
Tauchen (2005) builds a general equilibrium model with an endogenous volatility asymmetry. In
both models, the leverage effect is mainly determined by the relationship between volatility and
expected returns since neither of these theoretical models include corporate debt. We provide a
benchmark general equilibrium economy with debt to study the effects of financial leverage on the
1
dynamics of stock volatility.
Our work explores the dynamics of stock volatility at both a market and a firm level. We
characterize the economic channels through which financial leverage drives the dynamics of stock
volatility, and quantify the leverage effect. The assumptions in Black (1976) and Christie (1982) are
a constant asset return volatility and a constant riskless debt return. Both assumptions are inconsistent with realistic asset prices. We therefore study two different economies. In both economies,
the cash flows generated by a firm’s assets are specified exogenously, have a constant volatility, and
are split into an exogenously specified riskless debt service and a dividend stream to equity holders.
We derive the equilibrium prices and dynamics of all financial claims. We identify the economic
forces behind the dynamics of stock volatility and quantify the effect of financial leverage on the
dynamics of stock volatility.
The first economy we study is consistent with the assumptions in Black (1976) and Christie
(1982). Here, macroeconomic conditions are fixed and financial leverage is the only driving force
behind the dynamics of stock volatility. A firm’s financial leverage is solely driven by innovations
to the firm’s cash flows. Financial leverage generates little variation at the market level where
cash flow volatility is low, and significant variation at the firm level where cash flow volatility is
higher. When financial leverage is the only factor affecting the dynamics of stock volatility, the
leverage effect hypothesis holds at the firm level although stock volatility does not vary enough to
be consistent with empirical evidence.
The second economy has a representative agent with habit persistent preferences similar to
Campbell and Cochrane (1999). Such preferences lead to more realistic asset prices than in the
first economy. The driving force in this economy is counter-cyclical risk aversion caused by external
habit formation in the representative agent’s preferences. The model is calibrated to several features
of empirical asset prices, including the level and variation of the equity premium, the riskless rate,
and the market price of risk. We simulate the economy and explore the time-series behavior of a
firm’s stock returns and volatility allowing for both debt and equity. At both the daily and monthly
frequencies, we find at most a weak relationship between returns and volatility. We conclude
that financial leverage has little economic significance at the market level when macroeconomic
2
conditions lead to realistic dynamics for the riskless rate and the market price of risk. But at the
firm level, financial leverage can partially explain variations in stock return volatility.
2. Model
2.1. The Economy
We consider a continuous-time pure-exchange economy with a finite horizon T . Uncertainty in the
economy is represented by the filtered probability space (Ω, F, {Ft }, P ), with F ≡ FT , on which
is defined a two-dimensional Brownian motion W = (Wa , Wi ). The filtration {Ft } is generated
by W, augmented by the null sets. The uncorrelated components of the Brownian motion W
represent an aggregate consumption shock Wa and an idiosyncratic shock Wi . The idiosyncratic
shock is only used when studying the asset pricing of a small firm embedded in the economy.
All stochastic processes introduced are assumed to be progressively measurable, all (in)equalities
involving random variables hold P-a.s., and all stochastic differential equations are assumed to have
solutions without explicitly stating the required regularity.
The exogenous aggregate output in the economy δ(t) is assumed to follow a geometric Brownian
motion with dynamics given by
dδ(t) = δ(t) [µδ dt + σδ dWa (t)] ,
δ0 > 0,
(1)
where µδ and σδ are scalars.
Preferences are modeled in the economy by a representative agent with external habit formation
preferences as in Campbell and Cochrane (1999). The representative agent maximizes expected
lifetime utility given by
"Z
E
T
−ρt (C(t)
e
0
#
− X(t))1−γ
dt ,
1−γ
(2)
where C(t) is the representative agent’s consumption, X(t) is the habit level, γ captures the risk
aversion with respect to surplus consumption, and ρ is the subjective discount factor. When
X(t) = 0 for all t, the preferences are identical to a CRRA utility representation with coefficient of
3
relative risk aversion equal to γ.
In equilibrium, the representative investors consumes aggregate output:
C(t) = δ(t).
(3)
To complete our description of preferences, the process governing the habit X(t) must be defined.
Following Campbell and Cochrane (1999), we assume that the habit is external; agents do not
optimize over X(t). It is convenient to define the habit in terms of the surplus-consumption ratio
defined as
S(t) =
C(t) − X(t)
δ(t) − X(t)
=
,
δ(t)
δ(t)
(4)
with the second equality from the equilibrium condition (3).
The evolution of the habit level, X(t), is characterized by the mean-reverting log surplusconsumption ratio,
4
s(t) = ln(S(t)).
(5)
Unless noted otherwise, lower case letters denote the natural logarithms of the variables. The
dynamics of s(t) are
ds(t) = φ (s̄ − s(t)) dt + λ(s(t))σδ dWa (t),
s(0) ∈ (0, smax ),
(6)
where φ is the mean-reversion rate, s̄ is the steady-state level, and σδ is the volatility of the
aggregate consumption growth with S̄ and smax defined as
r
S̄ = σδ γ
1
,
φγ − B
smax = s̄ + 0.5 × 1 − S̄ 2 ,
(7)
(8)
with B a constant. As in Wachter (2006), our model is an extension of Campbell and Cochrane
(1999) to incorporate stochastic interest rates. The parameter B determines the sensitivity of
interest rates to the output shock. If B = 0, our model is the same as in Campbell and Cochrane
4
(1999) with constant interest rates.
The sensitivity of the surplus-consumption ratio to aggregate consumption shocks, λ(s), is
λ(s) =
1p
1 − 2 (s − s̄) − 1.
S̄
(9)
Given λ(s) > 0, the habit and aggregate consumption move together as any intuitive definition of
habit would require.
The surplus-consumption ratio S(t) captures the state of the economy. A low surplus-consumption
ratio implies that the economy has been hit by a succession of bad shocks; current consumption is
low compared to the historical average which is captured by the habit. Alternatively, a high S(t)
is an indicator of a good state of the economy, in which the agent enjoys higher consumption levels
compared to his habit. The local curvature of the utility function,
γ
S(t) ,
varies counter-cyclically
with S(t) and generates a counter-cyclical market price of risk in our model.
The representative investor finances his optimal consumption stream by trading in a locally
riskless money market with price B(t) and the market portfolio with price V (t). The market
portfolio is a claim to aggregate consumption; the cash flow from the market portfolio is split into
debt and equity. The money market is in zero net-supply, while the market portfolio is in positive
net-supply, normalized to one. The posited dynamics of the two securities are
dB(t) = B(t)r(t)dt,
B(0) > 0,
(10)
dV (t) + δ(t)dt = V (t) [µV (t)dt + σV,a (t)dWa (t)] ,
V (T ) = 0,
(11)
where the price system (r, µV , σV,a ) is determined in equilibrium.
Given our main goal is to explore the volatilities of debt and equity claims to V as well as
a small firm exposed to both aggregate and idiosyncratic risk, it is convenient to summarize the
endogenous price system in terms of a state price density process ξ(t) with dynamics given by
dξ(t) = −ξ(t) [r(t)dt + κ(t)dWa (t)] ,
5
ξ(0) = 1,
(12)
−1
where κ(t) , σV,t
[µV (t) − r(t)] is the market price of risk in the economy.
The representative agent’s complete markets dynamic opportunity set from trading the money
market and the market is summarized by a static budget constraint using standard martingale
techniques (Karatzas et al. (1987), Cox and Huang (1989)):
Z
T
ξ(t)C(t)dt ≤ V (0),
E
(13)
0
where the representative agent is initially endowed with the claim V .
We characterize asset prices using general equilibrium restrictions to identify the equilibrium
state price density process. The equilibrium consumption rules, asset prices, and asset holdings
must clear both the consumption and the financial markets, as well as satisfy the necessary and
sufficient conditions for optimality of the representative agent’s consumption-portfolio problem.
The martingale formulation of the representative agent’s optimal consumption-investment problem
provides us with a direct way to construct the equilibrium (Karatzas et al. (1990)). The state price
density process implied by the marginal utility of the representative agent evaluated at consumption
market clearing: C(t) = δ(t) gives equilibrium asset prices that clear the financial markets. The
equilibrium is characterized by solving for the unique state price density process. The necessary
and sufficient conditions for the representative agent’s optimization problem implies the equilibrium
state price density is
−ρt
ξt = e
S(t)C(t)
S(0)δ(0)
−γ
.
(14)
The endogenous equilibrium parameters r(t) and κ(t) are easily determined in terms of economic
primitives by computing the dynamics of equation (12),
σ2
r(t) = ρ + γ µδ − δ
2
−
κ(t) = γ (1 + λ(s(t))) σδ .
φγ − B
− B (s(t) − s̄) ,
2
(15)
(16)
In particular, r(t) and κ(t) both time vary as they are functions of the log surplus-consumption
ration s(t).
6
Additionally, the equilibrium state price density values any marketed consumption stream. In
particular, the value of the aggregate consumption stream is given by
1
E
V (t) =
ξ(t)
T
Z
ξ(s)δ(s)ds|Ft .
(17)
t
2.2. The Representative Firm
To evaluate the determinants of the return volatility at the aggregate level, we derive the equilibrium
dynamics of the market portfolio. In our model, the market portfolio is a representative firm that
pays out the aggregate consumption stream, δ(t), to its outside claimants. To explore leverage and
time-varying risk premium effects on aggregate volatility dynamics, the firm’s payouts are split the
into two financial claims.
The representative firm’s capital structure consists of debt and equity. All financial claims
issued by the firm are held by the representative agent in the economy. Since the representative
agent holds all claims to the firm in equilibrium, his incentives are aligned with total firm value
maximization. In such a first-best scenario, bankruptcy due to debt service payment defaults never
occurs along the equilibrium path, and the representative agent covers the debt service by injecting
capital into the firm when needed. This assumption can be justified by an infinitesimal, but positive,
bankruptcy cost. As a consequence, debt claims are not subject to default risk. Hence, the payout
claimed by the equity holders is equal to δE (t) = δ(t) − δD (t), where δD (t) is the required debt
service at time t. The specific terms of the debt service are described later. The total payout
generated by the firm is invariant to the capital structure of the firm. We abstract from frictions
such as security issue costs, taxation at both the corporate and personal level, and bankruptcy
costs.
The equilibrium price of the debt claim issued by the representative firm, D(t), is computed
using the equilibrium state price density (14), and is
1
D(t) =
E
ξ(t)
Z
Td
ξ(s)δD (s)ds|Ft ,
t
7
(18)
where Td ∈ [0, T ] is the maturity of the firm’s debt. The equilibrium debt dynamics are
dD(t) + δD (t)dt = D(t) [(r(t) + κ(t)σD,a (t)) dt + σD,a (t)dWa (t)] ,
(19)
for t ∈ [0, Td ] , with D(Td ) = 0, and where σD,a (t) is the endogenous debt return volatility. As
long as the equilibrium interest rate r(t) is time-varying, the volatility of the debt contract will be
non-zero even when the debt service δD (t) is deterministic as the contract is exposed to interest
rate risk.
The equilibrium price of the equity claim of the representative firm E(t) is
E(t) = V (t) − D(t).
(20)
dE(t) + δE (t)dt = E(t) [(r(t) + κ(t)σE,a (t)) dt + σE,a (t)dWa (t)] ,
(21)
Equity dynamics are
with E(T ) = 0, and where σE,a (t) is the endogenous equity return volatility.1
Equity is a claim to the residual cash flows given the current debt service structure. Since the
Miller and Modigliani Theorem (1958) holds in the model, all future capital structure adjustments
will be zero NPV transactions, and have no consequences on current equity prices and dynamics.
The next lemma formalizes the argument.
Lemma 2.1 Current equity prices and equity return dynamics are invariant to future capital structure adjustments.
The equilibrium dynamics of the debt and equity claims are fully characterized by equations (19)
and (21), once the conditional volatilities are determined. The procedure followed to numerically
obtain the endogenous asset dynamics is described in Appendix 6.
1
Throughout our analysis, we consider market-wide debt recapitalization policies that guarantee positive market
equity valuations.
8
2.3. A Small Firm
In addition to the representative firm, we also consider a small firm in the economy facing market
and idiosyncratic risk. We take the firm’s investment decisions as given, implying they are unaffected by its financing decisions. The firm is modeled as an EBIT-generating machine, with an
exogenous cash flow process
h
i
f
f
dδ f (t) = δ f (t) µfδ dt + σδ,a
dWa (t) + σδ,i
dWi (t) ,
δ f (0) > 0,
(22)
f
where µfδ is the instantaneous growth rate of the cash flows, σδ,a
is the factor loading on the
f
aggregate shock, and σδ,i
is the factor loading on the firm-specific shock. The total value of the
firm is
1
E
V (t) =
ξ(t)
f
T
Z
f
ξ(s)δ (s)ds|Ft .
(23)
t
The posited dynamics of the firm value process are:
i
h
f
f
(t)dWi (t) ,
(t)dWa (t) + σV,i
dV f (t) + δ f (t)dt = V f (t) µfV (t)dt + σV,a
(24)
with V f (T ) = 0.
The firm’s exogenous capital structure consists of debt and equity claims. The firm’s debt
f
(t) with maturity Td ∈ [0, T ]. The debt’s cash
service is a deterministic stream of cash flows δD
flows are riskless. To ensure this, we assume that the debt holders take control of the firm when
the value of the EBIT-generating machine falls to the debt’s value. At this point, debt holders can
costlessly replace their claim on the firm with a riskless security from the capital markets. Since the
equity holders do not have a bankruptcy timing option, the firm is able to issue debt with riskless
cash flows. However, the debt’s market value is risky due to its exposure to interest rate risk.
The equilibrium value of the firm’s outstanding debt is
1
D (t) =
E
ξ(t)
f
Z
Td
t
9
f
ξ(s)δD
(s)ds|Ft
,
(25)
with its equilibrium dynamics given by
f
dDf (t) + δD
(t)dt = Df (t)
h
i
f
f
f
r(t) + κ(t)σD,a
(t) dt + σD,a
(t)dWa (t) + σD,i
(t)dWi (t) ,
(26)
f
f
for t∈ [0, Td ] , with Df (Td ) = 0, and where σD,a
(t), and σD,i
(t) are endogenous factor loadings on
the aggregate and firm-specific shocks, respectively.
The firm’s equity is defined as a claim to the firm’s residual cash flow as long as the firm remains
solvent. Define the stopping time Tb :
Tb = inf{t : V f (t) = Df (t)}.
(27)
The value of equity is positive for t ≤ TB and is given by
E f (t) = V f (t) − Df (t),
(28)
for t ≤ TB and zero otherwise. Equilibrium dynamics are
f
dE f (t) + δE
(t)dt = E f (t)
h
i
f
f
f
r(t) + κ(t)σE,a
(t) dt + σE,a
(t)dWa (t) + σE,i
(t)dWi (t) ,
(29)
f
f
with E f (Tb ) = 0, and where σE,a
(t), and σE,i
(t) are the endogenous factor loadings on the aggregate
and the firm-specific shocks, respectively.
The equilibrium dynamics of the debt and equity claims are fully characterized by equations
(26) and (29), once the factor loadings are determined. A summary of the procedure used to obtain
the asset return dynamics is given in the Appendix.
3. Calibration and Simulation
We calibrate the exogenous model parameters to U.S. post-war data. The output process parameters are chosen to fit the annual growth rate of U.S. consumption data; we set µδ = 0.019 and
σδ = 0.015. The parameters that govern the dynamics of the surplus-consumption ratio are chosen
10
to fit the empirical moments of asset prices as summarized in the righthand column of Table 1.
The model is broadly consistent with the observed moments of asset prices for φ = 0.1385 and
B = 0.015. Other model parameters are specified as ρ = 0.09 and γ = 1.8. The first two columns
of Table 1 summarize the equilibrium asset pricing implications of our calibrated model. The first
column assumes their is no aggregate debt in the economy, while the second column assumes the
aggregate debt in the market takes the form of an annuity with a maturity of Td = 10 and a constant debt service of 60% of current total payouts. With such a capital structure, the unconditional
leverage ratio of the model roughly matches the U.S. mean aggregate leverage ratio from 1952-1998
as documented in Korajczk and Levy (2003).
Equations (9), (15), and (16) imply that the equilibrium riskless rate r(t) and the market price
of risk κ(t) are solely driven by the surplus-consumption ratio. Figure 1 plots the riskless rate, the
market price of risk, and the market’s price-dividend ratio in different states of the economy, as
well as the stationary distribution of S(t). Both the riskless rate and market price of risk move
counter-cyclically. At the extremely bad states of the economy, both variables rise dramatically.
The stochastic process of S(t); however, never attains a value of zero and the equilibrium asset
prices are well-defined in all states of the economy. Since the riskless rate fluctuates with the state of
the economy, all financial claims to future cash flows are subject to interest rate risk. Similarly, the
counter-cyclical variation in the market price of risk generates time-variation in the price-dividend
ratios of risky claims as can be seen in the bottom panel of Figure 1.
To explore the relationship between stock returns, stock volatility, and financial leverage, we
simulate the economy for 2,000 years at a daily frequency. Monthly data is then constructed
using the daily observations. The cash flows of the representative firm are given by the aggregate
consumption stream in the economy. We simulate two small firms with different cash flow risk
compositions. In particular, both small firms have identical expected cash flow growth rates µfδ =
f
0.019 and identical idiosyncratic cash flow risk volatilities σδ,i
= 0.2, whereas their aggregate cash
f
f
flow risk volatilities are different. We set σδ,a
= 0.015 for the firm we denote firm AI and σδ,a
= 0.0
for the firm we denote firm I. Therefore, the former has both systematic and idiosyncratic cash flow
risk, while the other firm has only idiosyncratic cash flow risk.
11
4. Results - Constant Asset Pricing Moments
Before exploring our fully calibrated economy, it is useful to consider a case fully consistent with
the leverage hypothesis where asset return volatilities are constant. Such a benchmark is easily
constructed by setting the habit level to zero or equivalently by setting φ = 0, S̄ = 1, and S0 = 1.
The model then collapses to the representative agent having standard power utility preferences.
With power utility preferences and i.i.d. consumption growth, the riskless rate and market price of
risk in the economy are both constant. This leads to unrealistic asset pricing implications, but is
consistent with the assumptions underlying the leverage effect hypothesis. This is a natural place
to start, and a useful benchmark for quantifying the leverage effect.
In this setting, closed-form solutions for the prices of all financial are easily obtained. Modeling
the debt service as a constant annuity of Td years with a constant cash flow equal to δD , the market
values of the representative firms’s total assets and debt are
h
i
δ(t)
−(r−µδ +γσδ2 )(T −t)
1
−
e
,
r − µδ + γσδ2
i
δD h
D(t) =
1 − e−r(Td −t) ,
r
V (t) =
where the constant riskless rate is given by r = ρ + γ µδ −
σδ2
2
−
γ 2 σδ2
2 .
(30)
(31)
The value of the firm’s
equity is then given by E(t) = V (t) − D(t) as long as E(t) ≥ 0.
For the representative firm, the debt volatility is zero since there is no interest rate risk. The
stock volatility is equal to
σE,a (t) =
D(t)
1+
E(t)
σδ .
(32)
A similar expression is obtained for the volatility of a small firm’s stock. Hence, in this economy,
stock volatility is driven only by the underlying cash flow volatility and the variation in the market
value of financial leverage.
To quantify the impact of financial leverage on the dynamics of stock volatility, we simulate our
economy.2 In our simulations for the levered aggregate firm, we assume that the firm rebalances
2
We use the same parameters as in the previous section with X(t) = 0 for all t.
12
its debt payments to a target level of 0.3 of total firm cash flows every 5 years. Table 2 shows
some descriptive statistics from the simulated market level data. When the representative firm is
unlevered, its stock volatility is identical to the dividend volatility, and shows no variation. When
the representative firm is levered, its stock volatility rises, and shows some variation over time.
With a realistic level of market leverage around 0.3; however, both the level and the variation in
market volatility is much lower than those observed in the data.
Table 3 and 4 provide similar descriptive statistics for the two small firms. Since both small
firms have larger cash flow volatilities than the representative firm, their stock return volatilities
are both higher on average. Since the debt’s value is constant for a given debt service, more volatile
stock returns lead to more volatile market leverage, which in turn generates further variation in
the stock return volatility.
While the CRRA preferences leads to a significant variation in stock volatility at the small firm
level, the stock volatility dynamics at the market level are quite unrealistic. Additionally, the model
does not generate a realistic spread between the cash flow volatility and the stock volatility.3 While
CRRA preferences imply a constant riskless rate and a constant price of risk, which are consistent
with the financial leverage hypothesis as in Christie (1982), several important characteristics of
asset prices are not obtained in this equilibrium.
5. Results - Time-Varying Asset Pricing Moments
We now calibrate our economy to the observed moments of asset prices. In contrast to the special
case studied in the previous subsection, the assumptions of the leverage effect hypothesis are not
satisfied in our calibrated economy. Because of the time-variation in riskless rate and market price
of risk, the value of a firm’s assets will have a time-varying volatility and the value of debt contracts
will be exposed to interest rate risk.
3
Shiller (1981) shows stock returns are at least five times more volatile than their underlying cash flows.
13
5.1. Market Level Analysis
We consider two different debt service structures for the representative firm: a 10-year constant
annuity and a constant perpetuity.4 While a 10-year debt maturity is realistic, the perpetual
maturity debt is widely used in the corporate finance literature for technical and computational
simplicity.5 Given we are not solving for optimal capital structures, our framework enables us to
analyze both cases and study the effect of debt maturity on equity return volatility. With either
maturity structure, we adjust the level of the coupon payment such that the debt claim comprises
30% of the total value of the firm initially.
The total value of the representative firm is driven by the instantaneous cash flows δ(t) and the
surplus-consumption ratio S(t). The value of the outstanding debt is similarly driven by δ d (t) and
S(t). Both the total firm value and the debt value are proportional to their underlying cash flows.
This implies that the equity value is proportional to its underlying cash flows and depends only on
δ e (t) and the surplus-consumption ratio S(t).
To see the impact of the macro factors on market volatility, Figure 4 plots market volatility at
different states of the economy keeping the ratio of debt service to total cash flows constant. The
figure makes clear that even the unlevered firm has significant variation in its stock volatility. In
good states of the economy, i.e. when S is high, stock volatility is low. As the riskless rate and the
market price of risk increases in the economy, stock volatility also rises. In extremely bad states of
the economy, the state variable S has very low volatility to keep it positive, leading to a decrease
in stock volatility. In most states of the economy, however, the inverse relationship between S and
stock volatility obtains.
While both the 10-year annuity and the perpetuity specifications are constructed with a 30%
financial leverage, their implications on the dynamics of market volatility are different. The impact
of financial leverage on stock volatility is higher when debt maturity is lower. The interest rate risk
exposure of the debt contracts depend on their duration. When debt has a low duration, its value
is less sensitive to increasing interest rates in bad times. Hence, market leverage increases faster as
4
In our analysis, we approximate a perpetuity by a 100-year constant annuity. Perturbing the maturity of this
annuity led to similar quantitative results.
5
See for example Fischer et al. (1989); Leland (1994); Leland and Toft (1996); Leland (1998); Goldstein et al.
(2001).
14
S decreases when debt duration is lower, leading to a higher stock volatility in bad states of the
economy.
We simulate our economy to see how financial leverage affects the variation in stock volatility.
We take the debt service as a 10-year annuity with an initial debt service to total cash flow ratio
of 74%, leading to 30% market leverage on average. We assume that the firm continuously issues
10-year discount bonds to keep the debt maturity constant at every point in time. We also assume
that the firm readjusts its debt service to 60% of the total cash flows every 5 years. Figure 5 plots
a 100 year window of the simulated market volatility and returns series.
Table 5 provides descriptive statistics on asset returns from the simulated data. Financial
leverage has a mean of 0.29 and a standard deviation of 0.06 for the market. When the market is
levered, both the mean and the standard deviation of realized returns increase. The average market
volatility rises from 13.95% to 19.01%, and the standard deviation of market volatility rises from
3.95% to 6.25%. Thus, financial leverage has increased both the level and the variation of market
volatility in our simulated data. The increased variation in market volatility, however, does not
necessarily imply that the leverage effect hypothesis holds true. The variation in financial leverage
that leads to variation in market volatility can also be driven by changing market conditions.
An implication of the leverage effect hypothesis is high realized returns are associated with low
market volatility. Thus, the probability density of realized returns should be negatively skewed. In
our simulated data, however, leverage leads to a small increase in the skewness of market returns.
The skewness of market returns increase from 0.07 to 0.09 in the daily data, and from 0.18 to 0.28 in
the monthly data. To see whether skewness is indeed increased by leverage, we plot the frequency
distributions of the volatility and returns. To visualize the impact of leverage on the skewness of
market returns, Figure 6 plots the frequency distributions of conditional volatility of equity returns,
realized returns, and expected returns, computed from the daily data. The figure makes clear that
there no observable change in skewness in market returns when leverage is introduced.
To further explore the relationship between market returns and market volatility caused by
leverage further, we replicate the regression exercises presented by the related empirical literature
using our simulated data. Following Christie (1982), Duffee (1995), and Schwert (1989), we consider
15
the following linear regression models:
σt+1
σt
= a0 + b0 Rt + t+1,0
(33)
log (σt+1 ) = a1 + b1 Rt + t+1,1
(34)
log (σt ) = a2 + b2 Rt + t,2
(35)
log
log (σt+1 ) = a3 + b3
Dt
+ t+1,3
Et
(36)
Dt
represents
Et
the market value of the firm’s leverage. Regression (33) was studied by Christie (1982) with b0 < 0
where σ represents the volatility of stock returns, R represents the equity return, and
consistent with the financial leverage hypothesis. Duffee (1995) argued that the source of a negative
coefficient on b0 was a positive contemporaneous relation between firm equity returns and firm
equity volatility by estimating the two regressions (34) and (35). Schwert (1989) estimated a
variation of regression (36); b3 > 0 is consistent with the financial leverage hypothesis.
To control for the time-variation in the market price of risk, we also consider an augmented
version of regression equation (36):
log (σt+1 ) = a4 + b4
Dt
+ c4 κt + t+1,4
Et
(37)
Table 6 summarizes the results of the regressions which should be interpreted as characterizing
the population moments of our model.6 In the daily data, the regression models (33), (34) and (35)
have no explanatory power regardless of the leverage. Model (36) has a high R2 and a significant
coefficient for market leverage. Model (37) makes clear that model (36) has an important omitted
variable: the market price of risk. When the market price of risk is introduced as an independent
variable, the sign of the coefficient of leverage changes. The results of the regressions using the
monthly data is similar except for the significant explanatory power of Model (33). The significant
negative relationship between realized returns and innovations in market volatility is consistent
with the financial leverage hypothesis. But, the financial leverage hypothesis implies that the b
6
We also ran kernel regressions to detect any non-linear relationships between returns and volatility. As no
significant non-linearities were detected, we only present the results for the linear regressions.
16
coefficient should be lower with increased financial leverage. The explanatory power in model (33)
is therefore not driven by effect of financial leverage.
Our analysis of the dynamics of market volatility shows that most of the variation in market
volatility is driven by the variation in market conditions. Financial leverage does not lead to
significant skewness in realized returns. The regressions we ran on our simulated data do not
provide support that financial leverage is a major driver of stock volatility.
5.2. Small Firm
While the aggregate consumption process restricts the calibration of the representative firm total
cash flow process, we are able to assume different risk loadings on the small firm’s cash flows. In
this section, we explore the impact of financial leverage on the dynamics of stock volatility at the
individual firm level, and in the presence of idiosyncratic cash flow risk.
We consider two different firms: a firm with both aggregate and idiosyncratic risk (firm AI),
and a firm with only idiosyncratic risk (firm I). For firm AI, we set σδ,a = 0.015 and σδ,i = 0.2,
while we set σδ,a = 0.0 and σδ,i = 0.2 for firm I. We assume that the levered firm restructures its
debt service to a target debt service to total cash flow ratio every 5 years.
We first analyze the equity return volatility dynamics of firm AI. Figure 7 plots the equity return
volatility in different states of the economy. The figure makes clear that the impact of financial
leverage on the dynamics of stock volatility depends on the firm’s debt duration. With shorter
duration debt, market leverage rises rapidly in the bad states of the world, leading to increased
equity volatility. For the long term debt, leverage increases the level of the equity volatility, but
leverage has little impact on the dynamics of the equity volatility.
Table 7 provides descriptive statistics on asset returns from the simulated data. For the levered
firm, financial leverage has a mean of 0.318 and a standard deviation of 0.117. Leverage increases
the average daily returns from 0.019% to 0.029%. Leverage increases the average equity volatility
from 24.47% to 37.02%. The impact of leverage on the variation in equity volatility is quite
significant. Leverage increases the standard deviation of equity volatility from 2.07% to 12.85%.
On the other hand, leverage increases the skewness of the equity returns. Figure 8 plots the
17
frequency distributions of equity volatility and returns, and clearly shows that leverage does not
induce negative skewness to equity returns.
Table 8 summarizes the results of the regressions for firm AI. We find no significant relationship
between realized returns and equity volatility in the daily data. In the monthly data, changes
in market volatility is negatively related to realized returns, as predicted by the leverage effect
hypothesis. The predictive power of this regression, model (33), is quite weak with an R2 of around
12%. Results of regressions (36) and (37) show that leverage has predictive power on stock volatility
even when the market price of risk is included in the regressions as an independent variable. Thus,
our regression analysis provide at least weak evidence supporting the leverage effect hypothesis.
Next, we analyze a small firm when all cash flow risk is idiosyncratic, firm I. Figure 9 plots
the equity return volatility in different states of the economy. Since all financial claims to the cash
flows of the firm are only subject to idiosyncratic shocks, the effect of changing market conditions
on equity volatility is only through the interest rate risk channel. As before, when debt has a low
duration relative to equity, market leverage increases rapidly in bad times, leading to higher stock
volatility.
Table 9 provides descriptive statistics on asset returns from the simulated data. As before,
leverage increases the level and the variation in stock volatility significantly. The variation in
volatility is lower compared to firm AI. The skewness of the return distribution does not seem to
decrease with leverage.
Table 10 summarizes the results of the regressions for the firm I. Model 33 provides no support
for the leverage effect hypothesis in both the daily and monthly data. Model 36 provides a significant
link between leverage and stock volatility. The predictive power of leverage on stock volatility is
strong even when the market price of risk is included as an independent variable.
Overall, our analysis provides some support that financial leverage drives the dynamics of stock
volatility at the firm level. This feature is driven by idiosyncratic risk influencing the firm’s equity
value and not the firm’s debt value. Hence, the firm’s financial leverage can move independent of
market conditions in contrast to our market-wide analysis. Time-varying market conditions are
still important determinants of even firm I’s equity volatility. Given the firm’s debt value is still
18
driven by systematic interest rate risk, variations in financial leverage are still partially explained
by systematic risk which ultimately feeds into the variation in the firm’s equity volatility.
6. Conclusion
We explore the relationship between financial leverage and the dynamics of stock volatility in an
economy with realistic interest rate and market price of risk dynamics. We show that for the
market as a whole, financial leverage increases the level of equity volatility, but the dynamics of
equity volatility are mostly driven by a time-varying interest rate and a time-varying market price
of risk. Financial leverage contributes more to the dynamics of stock volatility for a small firm
exposed to both idiosyncratic risk and market risk. But in both cases, the variation in interest
rates and the market price of risk is the main force behind the dynamics of stock volatility.
Understanding why stock volatility moves over time is crucial for many financial applications.
Equilibrium asset pricing models with realistic asset pricing implications can be used to explore the
dynamics of financial claims issued by firms. Our analysis relies on one of many such models, the
habit formation model of Campbell and Cochrane (1999), and assumes a simple riskless debt cash
flow structure. Future work is needed to extend this analysis to other asset pricing frameworks,
and explore the implications of risky debt cash flows to the dynamics of stock volatility.
19
References
Bekaert, G., and G. Wu, 2000, “Asymmetric Volatility and Risk in Equity Markets,” The Review
of Financial Studies, 13, 1–42.
Black, F., 1976, “Studies of Stock Price Volatility Changes,” Proceedings of the 1976 Meetings of
the American Statistical Association, Business and Economical Statistics Section, 177–181.
Campbell, J. Y., and J. H. Cochrane, 1999, “By Force of Habit: A Consumption-Based Explanation
of Aggregate Stock Market Behavior,” Journal of Political Economy, 107, 205–251.
Campbell, J. Y., and L. Hentschel, 1992, “No News is Good News: An Asymmetric Model of
Changing Volatility in Stock Returns,” Journal of Financial Economics, 31, 281–318.
Chen, L., P. Collin-Dufresne, and R. Goldstein, 2003, “Modeling the Dividend Process when Leverage Ratios are Stationary: Implications for General Equilbrium Models,” working paper, University of Minnesota.
Christie, A. A., 1982, “The Stochastic Behavior of Common Stock Variances - Value, Leverage,
and Interest Rate Effects,” Journal of Financial Economics, 10, 407–432.
Cosimano, T. F., Y. Chen, and A. A. Himonas, 2003, “By Force of Habit: An Exploration of Asset
Pricing Models Using Analytic Methods,” working paper, University of Notre Dame.
Cox, J. C., and C. F. Huang, 1989, “Optimal Consumption and Portfolio Policies when Asset Prices
Follow a Diffusion Process,” Journal of Economic Theory, 49, 33–83.
Duffee, G. R., 1995, “Stock Returns and Volatility : A Firm Level Analysis,” Journal of Financial
Economics, 37, 399–420.
Figlewski, S., and X. Wang, 2000, “Is the “Leverage Effect” a Leverage Effect?” Working Paper
New York University.
Fischer, E., R. Heinkel, and J. Zechner, 1989, “Dynamic Capital Structure Choice: Theory and
Tests,” Journal of Finance, 44, 19–40.
20
Goldstein, R., N. Ju, and H. E. Leland, 2001, “An EBIT-Based Model of Dynamic Capital Structure,” Journal of Business, 74, 483–512.
Karatzas, I., J. P. Lehoczky, and S. E. Shreve, 1987, “Optimal Portfolio and Consumption Decisions
for a ‘Small Investor’ on a Finite Horizon,” SIAM Journal of Control and Optimization, 25, 1157–
1186.
Karatzas, I., J. P. Lehoczky, and S. E. Shreve, 1990, “Existence and Uniqueness of Multi-Agent
Equilibrium in a Stochastic, Dynamic Consumption/Investment Model,” Mathematics of Operation Research, 15, 80–128.
Korajczk, R. A., and A. Levy, 2003, “Capital Structure Choice: Macroeconomic Conditions and
Financial Constraints,” Journal of Financial Economics, 68, 75–109.
Leland, H. E., 1994, “Corporate Debt Value, Bond Covenants, and Optimal Capital Structure,”
Journal of Finance, 49, 1213–1252.
Leland, H. E., 1998, “Agency Costs, Risk Management, and Capital Structure,” Journal of Finance,
53, 1213–1243.
Leland, H. E., and K. Toft, 1996, “Optimal Capital Structure, Endogenous Bankruptcy, and the
Term Structure of Capital Spreads,” Journal of Finance, 51, 987–1019.
Miller, M., and F. Modigliani, 1958, “The Cost of Capital, Corporation Finance, and the Theory
of Investment,” American Economic Review, 48, 261–297.
Schwert, W. G., 1989, “Why Does Stock Market Volatility Change over Time?” Journal of Finance,
44, 1115–1153.
Shiller, R. J., 1981, “Do Stock Prices Move Too Much to be Justified by Subsequent Changes in
Dividends?” The American Economic Review, 71(3), June, 421–436.
Tauchen, G., 2005, “Stochastic Volatility in General Equilibrium,” Working Paper Duke University.
Wachter, J., 2005, “Solving Models with External Habits,” Finance Research Letters, 2, 210–226.
21
Wachter, J., 2006, “A Consumption-Based Model of the Term Structure of Interest Rates,” Journal
of Financial Economics, 79, 365–399.
Wu, G., 2001, “The Determinants of Asymmetric Volatility,” The Review of Financial Studies, 14,
837–859.
22
Appendix
Computational Procedure
In this section, we describe the details of the numerical solution procedure used to compute the
asset return dynamics. We describe the details for the small firm analysis, since the market level
analysis is a special case with no idiosyncratic risk.
Since the cash flow growth is an i.i.d. process, the total firm value is characterized by two state
variables, i.e. V f (t) = V f (S(t), δ f (t)). A quick observation of equation (23) yields V (S(t), δ F (t)) =
δ F (t) × V (S(t), 1). Hence, to obtain the firm value, we first compute V (S(t), 1) for all S(t) ∈
(0, Smax ].
Numerically calculating V (S(t), 1) in the Campbell and Cochrane (1999) model is problematic; see for example Chen, Collin-Dufresne, and Goldstein (2003), Cosimano, Chen, and Himonas
(2003), and Wachter (2005). A convenient way to obtain V (S(t), 1) is to compute equation (23)
using a risk-neutral valuation. Under the risk-neutral measure, we can rewrite (23) as
Z
f
T
V (S(t), 1) = Ẽ
−
e
Rs
t
r(su )du f
δ (s)ds|Ft .
(38)
t
Then, by Monte Carlo simulating r(t) and δ f (s) under the risk-neutral measure, and using numerical
integration methods, we evaluate (38). Precise estimates are obtained by using 20,000 paths and
10,000 steps where T = 100.
To obtain the dynamics of the firm value, we apply Ito’s Lemma on V f (S(t), δ f (t)). The
diffusion terms of equation (24) are obtained as
f
σV,a
(t) =
S(t)λ(S(t)) ∂V f (S(t), 1)
f
σδ + σδ,a
,
∂S(t)
V f (S(t), 1)
f
f
σV,i
(t) = σδ,i
.
(39)
(40)
f
Given a deterministic debt service, the value of the debt claims are also given by D(S(t), δD
(t)) =
f
δD
(t) × D(S(t), 1). Hence, we obtain the debt value by computing D(S(t), 1) using Monte Carlo
simulations under the risk-neutral measure. The diffusion terms of equation (26) are obtained by
23
f
applying Ito’s Lemma on D(S(t), δD
(t)).
f
σD,a
(t) =
S(t)λ(S(t)) ∂Df (S(t), 1)
σδ ,
∂S(t)
Df (S(t), 1)
(41)
f
σD,i
(t) = 0.
(42)
f
As the residual claim, equity value is given by E(t) = δ f (t)V f (S(t)) − δD
(t)D(ST , 1). The
diffusion terms of equity return dynamics (29) are given by
f
f
f
σE,a
(t) = (1 + L(t)) σV,a
(t) − L(t)σD,a
(t),
(43)
f
f
σE,i
(t) = (1 + L(t)) σV,i
(t).
(44)
Proofs
Proof of Lemma 2.1.
Consider a firm whose assets pay the cash flow stream δ(t) and has a
required debt service δD (t). The value of the firm’s equity at time s is
1
E(s) =
E
ξ(s)
Z
T
ξ(t) (δ(t) − δD (t)) dt|Fs .
(45)
s
Consider a capital structure adjustment where the firm issues new debt contracts that require
a service δD (t) for t ∈ (Ta , Tb ], where s < Ta ≤ Tb ≤ T . The value of the newly issued debt Dn is
paid out to the shareholders on date Ta . The new equity price for the existing shareholders E n (t)
is given by
1
E(s) =
E
ξ(s)
+
+
1
E
ξ(s)
1
E
ξ(s)
Z
Ta
s
Z Tb
Ta
Z T
n
ξ(t) (δ(t) − δD (t)) dt + ξ(Ta )D (Ta )|Fs
n
ξ(t) (δ(t) − δD (t) − δD
(s)) dt|Fs
ξ(t) (δ(t) − δD (t)) dt|Fs .
Tb
24
(46)
(47)
(48)
Given no financial frictions when issuing the new debt, its value at time Ta is
1
D (Ta ) =
E
ξ(Ta )
n
Z
Tb
n
ξ(t)δD
(t)dt|FTa
.
(49)
Ta
Substituting (49) into (46) and applying the law of iterated expectations,
1
E(s) =
E
ξ(s)
Z
T
ξ(t) (δ(t) − δD (t)) dt|Fs .
s
implying that the equity price and dynamics are invariant to the future capital structure.
25
(50)
Table 1. Key properties of equilibrium asset prices. The levered market has a debt maturity of
Td = 10 and constant debt service at 60% of current total payouts.
Model (Unlevered)
Model (Levered)
Data
Expected Excess Market Returns
6.7%
8.5%
8.1%
Std. Deviation of Market Returns
13.7%
17.2 %
15.6 %
Expected Riskfree Rate
0.7%
0.7%
0.9%
Std. Deviation of Riskfree Rate
0.7%
0.7%
< 1.7%
Expected Market Price of Risk
0.43
0.43
0.39
0
0.30
0.30
Leverage
26
Table 2. Descriptive statistics of simulated equilibrium market returns with CRRA preferences.
Unlevered Market
Mean
Daily Returns
0.034%
Monthly Returns
1.040%
Conditional Volatility 1.500%
Std.Dev.
Minimum
Maximum
Skewness
Kurtosis
0.079%
0.436%
0.0%
-0.338%
-0.508%
1.500%
0.404%
3.055%
1.500%
0.006
0.008
N.A.
0.009
0.041
N.A.
Levered Market with Readjustment Every 5 Years
Daily Returns
Monthly Returns
Conditional Volatility
Market Leverage
Mean
Std.Dev.
Minimum
Maximum
Skewness
Kurtosis
0.035%
1.041%
2.106%
0.288
0.111%
0.613%
0.030%
0.010
-0.490%
-1.134%
1.997%
0.249
0.558%
3.830%
2.176%
0.311
0.006
-0.009
-0.490
-0.552
0.012
0.047
-0.312
-0.208
27
Table 3. Descriptive statistics of simulated equilibrium small firm with both systematic and
idiosyncratic cash flow risk equity returns with CRRA preferences.
Unlevered Market
Mean
Daily Returns
0.030%
Monthly Returns
0.905%
Conditional Volatility 20.06%
Std.Dev.
Minimum
Maximum
Skewness
Kurtosis
1.056%
5.886%
0.0%
-4.609%
-19.01%
20.06%
5.193%
24.99%
20.06%
0.033
0.171
N.A.
0.017
0.067
N.A.
Levered Market with Readjustment Every 5 Years
Daily Returns
Monthly Returns
Conditional Volatility
Market Leverage
Mean
Std.Dev.
Minimum
Maximum
Skewness
Kurtosis
0.028%
0.857%
31.42%
0.335
1.709%
9.508%
8.148%
0.113
-19.32%
-38.76%
22.53%
0.110
19.98%
74.56%
152.5%
0.868
0.047
0.326
3.837
1.131
1.817
1.762
24.36
1.457
28
Table 4. Descriptive statistics of simulated equilibrium small firm with fully idiosyncratic cash
flow risk equity returns with CRRA preferences.
Unlevered Market
Mean
Daily Returns
0.030%
Monthly Returns
0.902%
Conditional Volatility 20.00%
Std.Dev.
Minimum
Maximum
Skewness
Kurtosis
1.053%
5.871%
0.0%
-4.598%
-18.89%
20.00%
5.151%
24.73%
20.00%
0.032
0.171
N.A.
0.015
0.084
N.A.
Levered Market with Readjustment Every 5 Years
Daily Returns
Monthly Returns
Conditional Volatility
Market Leverage
Mean
Std.Dev.
Minimum
Maximum
Skewness
Kurtosis
0.028%
0.850%
31.44%
0.336
1.172%
9.548%
8.454%
0.114
-18.40%
-39.56%
22.54%
0.113
23.82%
74.65%
176.2%
0.886
0.055
0.344
4.127
1.137
2.150
2.081
28.66
1.504
29
Table 5. Descriptive statistics of simulated equilibrium market returns.
Unlevered Market
Mean
Daily Returns
0.021%
Monthly Returns
0.621%
Conditional Volatility 13.95%
Std.Dev.
Minimum
Maximum
Skewness
Kurtosis
0.764%
4.198%
3.953%
-4.110%
-17.21%
1.798%
4.769%
23.47%
18.31%
0.073
0.184
-0.836
0.678
0.692
-0.288
Levered Market with Readjustment Every 5 Years
Daily Returns
Monthly Returns
Conditional Volatility
Market Leverage
Mean
Std.Dev.
Minimum
Maximum
Skewness
Kurtosis
0.028%
0.849%
19.01%
0.289
1.054%
5.816%
6.254%
0.062
-6.099%
-25.50%
2.23%
0.188
7.285%
36.62%
28.84%
0.640
0.089
0.280
-0.448
1.299
0.961
1.043
-0.752
2.082
30
31
Model (37)
Model (36)
Model (35)
Model (34)
Model (33)
Model (37)
Model (36)
Model (35)
Model (34)
Model (33)
0.004
(0.0005)
−2.97
(0.003)
−2.97
(0.003)
−3.76
(0.006)
−3.39
(0.003)
a
−1.02 × 10−6
2.1 × 10−5
−1.73
(0.0005)
−1.73
(0.0005)
−2.54
(0.0001)
−2.15
(0.0005)
a
−0.46
(0.008)
−0.32
(0.05)
0.14
(0.045)
1.89
(0.013)
−2.80
(0.018)
0.92
0.43
6.8 × 10−5
6.8 × 10−5
−1.3 × 10−6
Adjusted R2
3.61
(0.013)
0.88
0.46
0.0004
0.002
0.124
Adjusted R2
Monthly Data
3.78
(0.002)
Levered Market
b
c
4.83 × 10−4
2.0 × 10−3
−0.33
(0.046)
−0.33
(0.046)
1.93
(0.0002)
−2.98
(0.003)
Levered Market
b
c
Daily Data
0.003
(0.0004)
−3.26
(0.002)
−3.26
(0.002)
a
−9.02 × 10−7
2.0 × 10−5
−2.02
(0.0004)
−2.02
(0.0004)
a
8.6 × 10−5
8.6 × 10−5
−1.3 × 10−6
−0.55
(0.01)
−0.44
(0.055)
0.11
(0.055)
0.0001
0.003
0.10
Unlevered Market
b
Adjusted R2
−6.16 × 10−4
2.6 × 10−3
−0.45
(0.06)
−0.45
(0.06)
Unlevered Market
b
Adjusted R2
Table 6. Estimates of the regression equations at the market level. Quantities in parenthesis are standard errors.
Table 7. Descriptive statistics of simulated asset prices for the small firm with systematic risk.
Unlevered
Daily Returns
Monthly Returns
Conditional Volatility
Mean
Std.Dev.
Minimum
Maximum
Skewness
Kurtosis
0.019%
0.565%
24.47%
1.295%
7.146%
2.068%
-6.062%
-23.90%
20.08%
6.632%
38.04%
27.11%
0.054
0.222
-0.387
0.105
0.091
-1.133
Levered with Readjustment Every 5 Years
Daily Returns
Monthly Returns
Conditional Volatility
Market Leverage
Mean
Std.Dev.
Minimum
Maximum
Skewness
Kurtosis
0.029%
0.847%
37.02%
0.318
2.068%
11.38%
12.85%
0.117
-62.31%
-70.49%
22.86%
0.091
41.25%
178.42%
491.96%
0.953
0.121
0.832
7.080
1.216
9.817
8.902
100.42
2.126
32
33
Model (37)
Model (36)
Model (35)
Model (34)
Model (33)
Model (37)
Model (36)
Model (35)
Model (34)
Model (33)
0.001
(0.0005)
−2.27
(0.002)
−2.27
(0.002)
−2.45
(0.002)
−2.64
(0.003)
a
−3.9 × 10−6
2.2 × 10−5
−1.03
(0.0004)
−1.03
(0.0004)
−1.25
(0.0003)
−1.42
(0.0003)
a
−0.18
(0.004)
−0.24
(0.02)
−0.06
(0.02)
0.33
(0.003)
0.27
(0.002)
0.88
0.75
0.0004
0.0004
2.7 × 10−4
Adjusted R2
0.54
(0.006)
0.73
0.56
0.001
0.01
0.12
Adjusted R2
Monthly Data
0.49
(0.0008)
Levered Firm
b
c
0.01 1.1 × 10−3
−0.24
(0.02)
−0.25
(0.02)
0.41
(0.0004)
0.34
(0.0003)
Levered Firm
b
c
Daily Data
2.8 × 10−4
1.2 × 10−4
−2.65
(0.001)
−2.65
(0.001)
a
−5.2 × 10−8
5.2 × 10−6
−1.41
(0.0001)
−1.41
(0.0001)
a
1.5 × 10−5
1.5 × 10−5
0.0
−0.05
(0.002)
−0.03
(0.011)
0.03
(0.011)
0.0003
0.0004
0.07
Unlevered Firm
b
Adjusted R2
−2.9 × 10−4
4.0 × 10−4
−0.03
(0.01)
−0.03
(0.01)
Unlevered Firm
b
Adjusted R2
Table 8. Estimates of the regression equations for the small firm with systematic risk. Quantities in parenthesis are standard
errors.
Table 9. Descriptive statistics of simulated asset prices for the small firm with no systematic
risk.
Unlevered
Daily Returns
Monthly Returns
Conditional Volatility
Mean
Std.Dev.
Minimum
Maximum
Skewness
Kurtosis
0.017%
0.498%
23.51%
1.243%
6.859%
1.815%
-5.656%
-23.22%
20.00%
6.294%
35.67%
25.95%
0.050
0.213
-0.279
0.093
0.076
-1.227
Levered with Readjustment Every 5 Years
Daily Returns
Monthly Returns
Conditional Volatility
Market Leverage
Mean
Std.Dev.
Minimum
Maximum
Skewness
Kurtosis
0.025%
0.732%
35.50%
0.319
1.961%
10.74%
11.32%
0.115
-55.38%
-77.40%
22.66%
0.092
26.78%
100.67%
470.42%
0.952
0.033
0.456
6.798
1.191
5.780
3.021
106.56
1.963
34
35
Model (37)
Model (36)
Model (35)
Model (34)
Model (33)
Model (37)
Model (36)
Model (35)
Model (34)
Model (33)
−2.2 × 10−6 −2.0 × 10−5
−1.07
(0.0004)
−1.07
(0.0004)
−1.30
(0.0003)
−2.68
(0.002)
a
−2.2 × 10−6
2.0 × 10−5
−1.07
(0.0004)
−1.07
(0.0004)
−1.30
(0.0003)
−1.45
0.0003
a
0.006
(0.001)
−0.25
(0.019)
−0.25
(0.019)
0.43
(0.0004)
0.36
(0.002)
0.89
0.78
0.0004
0.0004
9.0 × 10−5
Adjusted R2
0.42
(0.002)
0.82
0.78
0.0005
0.0004
0.0001
Adjusted R2
Monthly Data
0.42
(0.0007)
Levered Firm
b
c
6.0 × 10−3 1.0 × 10−3
−0.25
(0.019)
−0.25
(0.019)
0.43
(0.0004)
0.37
(0.0003)
Levered Firm
b
c
Daily Data
−5.5 × 10−8
4.6 × 10−6
−1.45
(0.0001)
−1.45
(0.0001)
a
−0.1 × 10−7
4.6 × 10−6
−1.45
(0.0001)
−1.45
(0.0001)
a
0.00001
0.00001
0.0
−2.8 × 10−4
3.7 × 10−4
−0.02
(0.01)
−0.02
(0.01)
4.4 × 10−6
4.7 × 10−6
0.0
Unlevered Firm
b
Adjusted R2
−2.8 × 10−4
3.8 × 10−4
−0.017
(0.011)
−0.017
(0.011)
Unlevered Firm
b
Adjusted R2
Table 10. Estimates of the regression equations for the small firm with no systematic risk. Quantities in parenthesis are standard
errors.
25
Probability Density
20
15
10
5
0
0
0.01
0.02
0.03
0.04
0.05
0.06
Surplus Consumption Ratio
0.07
0.08
0.09
0.1
0.14
0.12
Risk-Free Rate
0.1
0.08
0.06
0.04
0.02
0
-0.02
0
0.01
0.02
0.03
0.04
0.05
0.06
Surplus Consumption Ratio
0.07
0.08
0.09
0.1
0
0.01
0.02
0.03
0.04
0.05
0.06
Surplus Consumption Ratio
0.07
0.08
0.09
0.1
0.07
0.08
0.09
0.1
2.5
Price of Risk
2
1.5
1
0.5
0
P/D Ratio (Consumption Claim)
35
30
25
20
15
10
5
0
0.01
0.02
0.03
0.04
0.05
0.06
Surplus Consumption Ratio
Figure 1. Equilibrium properties of the pricing kernel and the market portfolio’s price-dividend
ratio at different states of the economy.
36
P/D Ratio (Consumption Claim)
35
30
25
20
15
10
5
0
0.01
0.02
0.03
0.04
0.05
0.06
Surplus Consumption Ratio
0.07
0.08
0.09
0.1
Figure 2. The dividend-price ratio of the market portfolio in different states of the economy.
0.7
10-year Maturity
Perpetuity
0.65
60
0.6
0.5
40
0.45
0.4
20
0.35
0.3
0.25
0
0.2
0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
0.09
0.1
Surplus Consumption Ratio
Figure 3. Market leverage for two different debt service maturity specifications.
37
Difference (%)
Market Leverage
0.55
10-Year Constant Annuity
Unlevered Market
Levered Market
Difference(%)
30
60
Volatility (%)
20
40
15
Difference (%)
25
10
20
5
0
0
0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
0.09
0.1
Surplus Consumption Ratio
Constant Perpetuity
Unlevered Market
Levered Market
Difference(%)
30
60
Volatility (%)
20
40
15
Difference (%)
25
10
20
5
0
0
0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
0.09
0.1
Surplus Consumption Ratio
Figure 4. Conditional volatility of equity returns of the representative firm at different states of the
economy, for a given debt service structure.
38
0.02
0.04
0.06
0.08
Conditional Return Volatility
100
120
140
160
180
200
180
200
180
200
180
200
0.000
0.010
0.020
0.030
Expected Returns
100
120
140
160
−0.2
0.0
0.2
Realized Returns
100
120
140
160
−0.2
0.0
0.2
Realized Excess Returns
100
120
140
160
Figure 5. A 100 year window of observations from the simulated data at the market level.
39
30
25
20
15
10
Probability Density
5
0
0.00
0.05
0.10
0.15
0.20
0.25
0.30
6
4
2
0
Probability Density
8
Volatility
0.0
0.1
0.2
0.3
0.4
50
40
30
20
10
0
Probability Density
60
Expected Returns
−0.04
−0.02
0.00
0.02
0.04
0.02
0.04
50
30
0
10
Probability Density
70
Realized Returns
−0.04
−0.02
0.00
Realized Excess Returns
Figure 6. Probability density estimates of daily observations from the simulated data. The solid
line denotes an unlevered economy. The dashed line denotes an economy with a 10 year annuity
debt structure. The dash-dot line denotes an economy with a perpetual debt structure.
40
10-Year Constant Annuity
60
Unlevered Market
Levered Market
Difference(%)
40
40
30
25
Difference (%)
Volatility (%)
35
20
20
15
0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
0.09
0.1
Surplus-Consumption Ratio
Constant Perpetuity
60
Unlevered Market
Levered Market
Difference(%)
40
40
30
25
Difference (%)
Volatility (%)
35
20
20
15
0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
0.09
0.1
Surplus-Consumption Ratio
Figure 7. Conditional volatility of equity returns of the small firm with both systematic and idiosyncratic cash flow risk at different states of the economy, for a given debt service structure.
41
40
30
20
10
0
Probability Density
Unlevered
Levered C.A.
Levered P.A.
0.20
0.25
0.30
0.35
0.40
0.45
0.50
5
4
3
2
0
1
Probability Density
6
Volatility
0.0
0.1
0.2
0.3
0.4
0.5
30
20
10
0
Probability Density
40
Expected Returns
−0.04
−0.02
0.00
0.02
0.04
0.02
0.04
30
20
10
0
Probability Density
40
Realized Returns
−0.04
−0.02
0.00
Realized Excess Returns
Figure 8. Probability density estimates of daily observations from the simulated data for the
small firm with both systematic and idiosyncratic cash flow risk.
42
10-Year Constant Annuity
60
Unlevered Market
Levered Market
Difference(%)
40
40
30
25
Difference (%)
Volatility (%)
35
20
20
15
0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
0.09
0.1
Surplus-Consumption Ratio
Constant Perpetuity
60
Unlevered Market
Levered Market
Difference(%)
40
40
30
25
Difference (%)
Volatility (%)
35
20
20
15
0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
0.09
0.1
Surplus-Consumption Ratio
Figure 9. Conditional volatility of equity returns of the small firm with idiosyncratic cash flow risk
at different states of the economy, for a given debt service structure.
43
40
30
20
10
0
Probability Density
Unlevered
Levered C.A.
Levered P.A.
0.20
0.25
0.30
0.35
0.40
0.45
0.50
5
4
3
2
0
1
Probability Density
6
Volatility
0.0
0.1
0.2
0.3
0.4
0.5
30
20
10
0
Probability Density
40
Expected Returns
−0.04
−0.02
0.00
0.02
0.04
0.02
0.04
30
20
10
0
Probability Density
40
Realized Returns
−0.04
−0.02
0.00
Realized Excess Returns
Figure 10. Probability density estimates of daily observations from the simulated data for the
small firm with fully idiosyncratic cash flow risk.
44